Humans vary from one another by single nucleotide polymorphisms (SNPs), with 10 million SNPs being common (allele frequency >1%). Only a small fraction of SNPs are thought to have functional or phenotypic consequences. SNPs associated with diseases are often functionally neutral, correlated with a particular phenotype only due to linkage disequilibrium. By contrast, insight into disease mechanism is primarily provided by knowledge of the functional variant. Systematically relating functional genetic variation to disease and drug response is a major goal of the Genes, Environment and Health Initiative. Known disease-causing mutations are overwhelmingly coding and measures of selection show a significant fraction of non-synonymous SNPs (nsSNPs) to be evolutionarily deleterious. Nevertheless, the majority of nsSNPs probably have limited effect on molecular function and phenotype. Several methods have been developed to discriminate functional from non- functional nsSNPs, exploiting a combination of sequence and evolutionary information. Although, these predictions can serve as a guide, they are not currently a substitute for direct experiment. Here we investigate the effects of human coding variants on protein kinase function. Protein kinases are well studied and have been associated, often causally, with many diseases. Mutations that inappropriately activate kinases can drive the development of diseases such as cancer. In preliminary work we have learned that exogenous expression of protein kinases can drug response in cultured human cells dramatically. Our first goal will be to generate a resource of ~3000 open reading frames capturing variants of 535 human protein kinases. All clones will be cloned into lentiviral expression vectors to allow transfection into a wide range of human cells and tissues. This resource will contain wild type and a deep sampling of variants, with emphasis on common and predicted-deleterious variants. Each clone will sequence verified and uniquely bar-coded. Our second major goal will be to determine how expression of kinases and their variants modify drug response. Kinase expression can quantitatively and reproducibly enhance or block the action of drugs in tissue culture. For example, we have shown that response to the EGFR drug Iressa is blocked by overexpression of a number of kinases including those that act downstream in EGFR pathway. We will use functional tests to determine how kinase variants change response to three common cancer chemotherapy drugs. Once established, the proposed framework can be economically adapted to the action of any drug with a quantitative phenotype in cultured human cells and general to other classes of genes and coding variants. Genetic variation is a major but still poorly understood contributor to human health, altering disease susceptibility, resistance, and response to treatment. In this application we test how sequence variation in one important class of genes, the protein kinases, and affects response to important anti-cancer drugs. We pilot an economical large-scale framework for relating individual genetic differences to therapeutic drug response, while offering an experimental avenue to study any drug that induces a quantifiable response in cultured human cells. We will construct a 3000-clone expression library representing variants of 535 protein kinases in a bar-coded lentivirus backbone that will be of high value to the research community, and we will generate a large set of previously unobtainable data on kinase-dependent regulation of human cellular metabolism.